Detection of objects including persons using image processing

Abstract

We studied methods for observing objects including persons. First, we clarified the limit of exponential smoothing using fixed point operations executed by general image processors, then we proposed a method for detecting objects that are left in a crowd using exponential smoothing with accumulation of binarized images. The experimental results of left objects detection had a higher than 90% recognition rate. Finally, we proposed a doubtful objects detection method by analyzing objects movements. We clarified the suitability of the proposed method by experiments.